Socure: identity verification, fraud decisions and RiskOS operations

Socure supplies identity and fraud technology to financial-services businesses and other organisations. RiskOS brings checks, risk signals and decision workflows into a shared operating environment. For a broker, the practical connection is the customer lifecycle: establish who is opening an account, assess suspicious applications, verify a funding account and route exceptions to reviewers.[1]

Socure Launch is the self-service entry path with preconfigured solutions. Enterprise supports customised orchestration, advanced integrations and a tailored rollout. Both use RiskOS, but the flexibility and limits of one package should not be transferred to the other.[2]

RiskOS can return ACCEPT, REJECT or REVIEW through configured workflow logic. Manual-review paths create cases for human decisions, which are then communicated through a webhook or polling. The distinction matters operationally: a workflow can automate a customer’s decision policy, while an explanation, risk score and manual-review outcome remain different outputs.[9]

A hosted flow lets Socure provide the customer-facing collection interface; a direct API integration uses the broker’s own frontend. Both connect to the workflow engine. Device collection still needs the relevant SDK and session-token handoff rather than a server request alone.[2][8]

The MCP server can expose documentation, schemas and workflow metadata to development assistants, and its published tool list includes creating and updating webhooks. It does not execute transactions or evaluations. Access uses a separate MCP Server Key; username/password authentication ceased after 21 July 2026. Its documented write authority concerns webhook configuration.[12]

Launch publishes the following production evaluation prices and $1,000 in monthly usage credits. These are the dollar amounts printed on the public page; it does not identify an explicit currency code.[13]

Enterprise pricing is customised to the deployment.[2]

Launch bills each initiated production evaluation, including abandoned workflows. It permits one solution, one evaluation per second and 500 daily; support handles package and integration-mode changes. Solutions labelled for US evaluations target US individuals; non-US submissions can be incomplete or return no decision.[14]

Account Intelligence separately advertises pay-per-hit charging for successful account verifications, without a numerical unit tariff. That product-specific basis should not be confused with Launch’s initiated-evaluation charge.[5]

Socure’s go-live checklist calls for correct field mapping, tested webhook and case flows, and at least three evaluation IDs for workflow certification. SDK deployments also involve a Developer Support Engineer. It requests at least one week’s notice for elevated throughput. Its separate reference to a 10 TPS default does not establish that an unactivated production account is usable.[18]

Access design can separate reviewing an existing case from initiating or rerunning an evaluation. In the documented client-account role model, Can Review controls decisions, while Can Evaluate creates a fresh risk evaluation. Global custom-role edits propagate immediately across client accounts; local roles stay within the selected client. Permissions exposed in the dialog depend on configuration and enabled products.[20]

The security guide recommends TLS 1.3, with TLS 1.2 as the minimum, and describes MFA, SAML sign-in and role-based controls. IP allowlisting is optional. These controls complement the broker’s responsibility for keeping backend secrets out of frontend code and restricting credential access.[19]

Enterprise includes dedicated support, without a public universal incident-response or restoration timetable. Launch support handles account creation and package or integration-mode changes.[2][14]

Socure describes its identity-verification and fraud products as outside consumer-reporting use and says its contracts prohibit using them for FCRA eligibility decisions. An identity-fraud threshold can inform onboarding, but the documentation distinguishes that process from assessing creditworthiness. It also requires customers to present the CRAIN notice before applying services to UK individuals. These are product-use conditions, not a financial licence for the broker.[23]

DocV’s notice and consent screen cannot be removed, customised or disabled in the native or web capture application. Declining consent terminates capture. After consent, an unfinished capture expires after one hour and must be restarted; a web implementation needs a new link. The platform can export a consent audit report showing agreement or refusal.[24]

The 1 October 2026 privacy notice distinguishes customer-instructed processing from Socure’s controller activities, including DocV and derived fraud intelligence. Identity images may support model training and fraud records; this is not a blanket no-training service. The notice describes explicit consent for relevant UK/EEA biometric and training processing.[25]

The notice identifies US storage, with service-provider processing and staff access in additional countries.[25]

The service operator identified in the privacy notice is Socure Inc.[25]

The acquisition of Effectiv expanded Socure’s decisioning and orchestration capabilities beyond identity checks into connected onboarding, authentication, payment-fraud and transaction-monitoring workflows. The acquisition material also describes integrating third-party data alongside Socure products. RiskOS can therefore combine Socure checks with separately sourced enrichments in one decision process.[26]

On 27 August 2026, Socure announced a growth investment at a $5.2 billion valuation led by Summit Partners, with Goldman Sachs Alternatives, Wells Fargo and Docusign among participants. It also announced acquiring Fravity and plans to integrate that business’s agent-building capabilities into RiskOS. Fravity integration was presented as intended development in that announcement.[27]

Operator

Socure Inc.

Service scope

Identity, document and biometric verification, fraud analysis and decision orchestration.

Customer types

Financial-services businesses; Fintechs; Banks; Other organisations

Delivery

Hosted; API

Pricing publication

Public and quoted pricing

Platforms

RiskOS; Socure Launch; Socure Enterprise

Integrations

Evaluation API; Digital Intelligence SDKs; DocV SDKs; Webhooks; Okta identity proofing; MCP server

Pricing scope

Launch publishes per-evaluation prices; Enterprise pricing is customised.

Implementation

Hosted or direct API integration, configured workflows, testing and production activation.

Support

Enterprise dedicated support; Launch changes and technical enquiries use Socure Support.

Permissions and regulatory scope

Identity and fraud technology with documented product-use restrictions; customer decisions and obligations remain distinct.

Provider metadata checked 20 Oct 2026Source 1Source 25Source 2Source 13Source 23Source 10Source 12Source 8

The included GenAI Explainability feature creates a saved, plain-language explanation inside Case Management. Socure says it does not change the decision and runs inside its environment. The feature is enabled by default, can be disabled through support and requires case-view access. The current guide provides neither an API nor Sandbox availability. Its statement about the explanation environment should be read separately from the broader privacy notice’s other model-training uses.[28]

The 6 October 2026 DocV Omega update adds activeModelScore while preserving the existing decision and reason-code logic. A new score and the legacy decision can therefore point to different outcomes. Resubmission transactions are excluded from Omega scoring. For a broker changing thresholds, that is a version-specific output distinction to preserve in the integration, rather than treating every returned field as the same decision.[29]

Watchlist monitoring uses the screening enrichment’s referenceId, not the top-level eval_id. A workflow can enrol only approved customers; the API and batch files can enable or disable monitoring separately. The auto-monitoring switch instead enrols all evaluated entities and is described as useful during an existing-customer import, with unapproved entities subsequently removable. This affects both migration design and who remains under ongoing observation.[7]

Socure’s go-live guidance explicitly says RiskOS is not the customer’s long-term system of record and links retention to the contract. It recommends retaining evaluation IDs, scores and reason codes locally. The public export tools describe available records, but the reviewed documentation does not establish a universal post-termination retrieval period or enterprise deletion deadline.[18]